Online Temperature Modeling in the Hangzone Sonnenberg, Zurich ^†

Abstract

Continuous measurements within the water distribution network of the Water Supply Zurich (WVZ) revealed surprisingly high water temperatures in different locations, especially during the summer months. Initial investigations were unable to determine the source of the elevated water temperatures, so the WVZ initiated a pilot project for online monitoring and modeling of the water temperature in the network for the Hangzone Sonnenberg pressure zone. Modeling the interaction of the initial water source temperature, water source mixing and heat exchange due to ground temperature is one of the main challenges in this project.

1. Introduction

Climate measurements show an increasing long-term trend of the mean air temperature [1], which has a direct impact on the ground temperature, and the latter influences the water temperature in the pipes. Measurements directly in the pipe network of Zurich revealed high water temperatures, especially during the summer, approaching the limit of 25 °C [2] given by the Swiss regulations [3]. An online model can help to monitor the network to identify critical locations contributing to the elevated water temperatures and to think about possible actions to mitigate the heating effect.

The chosen pressure zone is the Hangzone (HZ) Sonnenberg, consisting of 22,000 pipes, 2 reservoirs and 7 pumps (Figure 1a). In this pressure zone, water from the sources is first pumped into a tunnel and then delivered into the network through different outlets. The HZ Sonnenberg has two main water sources named Hardhof and Lengg. From Hardhof, groundwater is pumped with typical temperatures around 12–18 °C, whereas Lengg provides lake water with temperatures around 6–10 °C (30 m under the water surface).

The online model is gradually being updated and calibrated as more flow, pressure and temperature measurements are installed. By now, 279 online measurements are incorporated and some are used as boundary conditions for the online model.

2. Online Temperature Modeling

2.1. Theoretical Background

Temperature modeling considers hydraulic advection and mixing processes and it also considers heat transfer through conduction from the pipe material. The latter depends on the thermal conductivity of the material, the difference in temperature between the water and the pipe and the area of contact between the two objects.

In case of a water pipe, the thermal conduction process will consider the pipe diameter, material and wall thickness, and also the flow velocity and the surrounding soil type. Measurements have shown that the water temperature undergoes either heating or cooling effects, approaching the limiting ground temperature. Depending on the different parameters and properties, this can happen within a distance of 500 m or within a few hours [4].

2.2. Practical Approach

The simulation of the water temperature was added into the DHI EPANET hydraulic solver using the following formula:

T(t) = T_e + (T₀ − T_e) e^−kt,

(1)

where

• T(t) is the temperature of a water parcel at a time step t;
• T_e is the constant temperature of the environment (ambient temperature);
• T₀ is the initial temperature of the water parcel;
• k is a constant depending on the object’s properties (bulk coefficient of heat transfer reflecting specific heat capacity, mass flow between the pipeline and surroundings).

Experimental observations would be needed to determine the heat transfer coefficients across different parts of the distribution network. With respect to practical aspects of using this concept in the water temperature modeling, these coefficients were estimated and adjusted to match the water temperature as measured in selected locations.

The water temperature modeling runs as part of the real-time hydraulic model and it is updated every 10 min. Water temperature results are extracted from the result files and stored in the MIKE WaterNet Advisor database for archiving and display purposes as well as for comparison with water temperatures measured at selected locations.

3. Status of Temperature Model

3.1. Boundary Conditions

The two sources Hardhof and Lengg are used as boundary conditions for the temperature, together with an input of the yearly time series of the ground temperature. The latter was built based on past observations from two different measurement stations nearby, one under asphalt and one under grass, both at 1.6 m depth (pipe depth). When looking at the entire pressure zone this is a generalization, because the ground properties in different parts of the city can vary significantly. This could lead to errors in the modeled temperatures.

3.2. Calibration Status

The basic set of parameters in the EPANET quality module to model the temperature includes, among others, the bulk reaction order (first order), and the bulk coefficient. The simplification of using a global bulk coefficient for the entire network could not bring satisfying results. The evaluation of the heat transfer is performed mostly in the north-western and south-eastern parts of the network. There, the water is relatively homogeneous, i.e., mixing of source waters occurs rarely or never. In these regions, close to the source, it is also easier to determine the flow path. This allows to adjust the bulk coefficients to match the modeled temperature with the measurements. Therefore, each pipe was assigned with a different coefficient depending on its diameter: the bigger the pipe, the lower the bulk coefficient, so the heat exchange happens slower. The definition of different bulk coefficients still needs to be optimized.

3.3. Results

The results showed that the heating in the network, due to the ground temperature, needs to be accelerated in smaller pipes and slowed down in bigger pipes. Figure 1 shows results from four different stations along a flow path:

• Lengg: source, starting point;
• Burgwies: 1 km in the DN 1200 mm tunnel;
• Vogelsangstrasse: another 2 km in the DN 1200–1350 mm tunnel and then 2 km in a DN 550 mm pipe;
• Sonneggstrasse: from different parts of the network, mainly in DN 150–200 mm pipes.

4. Next Steps and Main Challenges

There are several issues improving the results of the online temperature model.

4.1. Ground Temperature

Measurements in the network confirmed that the ground temperature plays a key role when modeling water temperature online. Historical observations of two stations nearby are used as inputs. The spatial variability of the ground temperature throughout the network is not taken into consideration yet. This can only happen by enlarging the measuring network and classifying different ground types and properties all over the city. Online measurements of the ground temperature would improve the model results.

4.2. Bulk Coefficients

The bulk coefficient for the heat exchange rate depends on both the pipe diameter and the pipe material. On one hand, the pipe diameters typically decrease from the source to a demand point in the network. This behavior helps in tuning the bulk coefficients during calibration because the effect of changes can be seen better. On the other hand, there is no typical sequence of pipe materials along a flow path. This means that the calibration would be more complicated considering that the bulk coefficient should be dependent on both diameter and material at the same time.

4.3. Adding a New Limiting Temperature

The tunnel is buried up to 100 m deep in the ground, much deeper than the normal pipes (1.5 m). A second limiting ground temperature is needed to model the heat processes in the tunnel, where the temperature time series from 1.6 m depth cannot be applied. Instead, a constant temperature must be assumed at the tunnel depth [5]. This temperature cannot be measured but can be calibrated using measurements of stations in some tunnel outlets.

4.4. Possible Improvements in the Model Implementation

Maintaining the hydraulic calibration of the model is a basic precondition of trustful results of the water quality modeling. Therefore, the online model inputs incorporate the SCADA data and a pipe shut-down register. In addition, the model is updated from the GIS system on a regular basis and recalibrated after significant changes in the operation scheme. Increased measurements in the network are not only an incentive for model recalibration, but have also facilitated dividing the current zone into three parts, with individual water balances for each part.

5. Practical Usage of the Temperature Online Model

The online model has already been used in the past for the optimal location of pressure, flow and water quality measurement profiles and it is expected that, with the increasing knowledge of water temperature behavior and critical locations, these measurements will be further complemented. The model will be used to investigate possibilities to influence the temperature and avoid or mitigate negative impacts of exceeding temperature limits of the delivered water. In particular, the impact of changing the water distribution scheme (e.g., by closing valves) will be investigated.

6. Conclusions

Water temperature measurements within the distribution network show high temperatures which correspond well with measured ground temperatures. The online model has demonstrated potential to help the WVZ evaluate strategies to mitigate the high water temperatures like planting trees, using shaded areas or bright asphalt, implementing the concept of a “sponge-city”, and so on. However, the actual temperature online model still has its limitations and uncertainties that hinder the daily usage of the system as a decision tool. The recommended improvements described in chapter 4 are expected to address these limitations and significantly improve the performance and utility of the online model.